Office of Graduate Studies and Postdoctoral Affairs

Draft Summary Paper of Big Data Fellowships, prepared by Laura Alexander and Phil Trella

Big Data Fellowships in 2013-14 and 2014-15

Through a grant from the Jefferson Trust Foundation and in collaboration with the newly founded Data Science Institute, the Office of the Vice President for Researchhas been working to create opportunities for graduate students in diverse disciplines to work together on collaborative research projects in the area of Big Data. The following summary details these efforts over the course of the last two years, culminating in recently announced awards to five new collaborative research teams.

Beginning with an initialRequest for Applications in the spring of 2013, students across the university were invited to find collaborators in other disciplines and to submit a joint research proposal in the broad area of Big Data, requiring the disciplinary expertise of each applicant. The terms of the fellowship were established as one year of financial support for each graduate student collaborator in the amount of $20,000 with an additional $5000 for an undergraduate student collaborator. Altogether, we received five joint proposals, which were reviewed by an interdisciplinary committee of faculty. Three were chosen for funding, ultimately resulting in eight graduate student fellowships, with one undergraduate assistant, for the 2013-2014 academic year. Below are the projects/students funded:

  • Replacing the Bar Graph: New Approaches to Big Data Visualization: William Green and Elizabeth O’Brien, Architecture, and Joseph Walpole, Biomedical Engineering, assisted by undergraduate Angela Jividen, developed new ways to visualize large sets of data, in order better to convey information about biological processes to enhance the understanding and work of researchers who are developing new treatments for disease and mechanisms for improving health.
  • Counting Bicycles Leveraging Existing Sensors:Conrad “Alec” Gosse (PhD 2014), and Han Du in Civil and Environmental Engineering, teamed up with Emmanuel Denloye-Ito in Electrical and Computer Engineering,to develop a system for tracking traffic patterns in cities – including bicycle and pedestrian patterns – in order to better inform urban planning around bike use.
  • MOOCs as a Massive Laboratory: Ignacio Martinez, Economics, and Paul Diver, Statistics, studied student participation and success rates in Massive Open Online Courses, in order better to understand student learning patterns and how to encourage success in learning, both online and off.

In the process of advertising the RFP and making decisions on the first round of proposals we received excellent feedback from both faculty and students, which helped to shape the process that we put into place over the course of 2013-14.Most of this feedback highlighted the need to create more “scaffolding” surrounding the program, including, in sequence:

  1. More opportunities for students to formulate research questions in the area of “Big Data” within their own disciplinary frameworks.
  2. Additional opportunities, incentives, and guidance for students to find collaborators for these ideas, and to mesh them into a sound, collaborative research proposal.

In order to address this feedback over the course of 2013-2014 the process for facilitating collaborations and soliciting proposals proceeded through several successive stages, designed to stimulate interest and offer opportunities and guidance as students moved their ideas from conception, to finding collaborators, and eventually to the formulation of a well-designed collaborative research proposal.

Beginning in October 2013 we disseminated specialized invitations to each school/department on Grounds offering to sponsor a departmental roundtable discussion specifically for graduate students to brainstorm ideas, from their own disciplinary perspectives, on how they could collaborate with other disciplines on research problems surrounding Big Data. Students were invited to plan and organize their own departments’ gatherings, and we reimbursed students for food served at the meetings. We also provided students, if requested, with broad questions they might use to spark discussion. Discussions were intended to be wide-ranging and to fit the interests and ideas of each department’s students, so we provided only minimal direction to student facilitators. Departments that reported participation included Economics, Biology, Philosophy, Biomedical Engineering, Politics, Systems Engineering, Statistics, Music, and Astronomy. At least 70 students participated in these roundtables,with 80-90 participants the most likely number.

The promotion and sponsorship of the department-based roundtable discussions was designed to culminate in an event for the expressed purpose of facilitating connections between potential collaborators across Grounds. The “Big Data Mixer,” which took place on November 20, 2013, was attended by 60 students representing a wide variety of disciplines in social sciences, humanities, engineering, biosciences, physical sciences, nursing, languages and architecture.

In order to facilitate students getting to know each other’s work and brainstorming collaborations, the mixer began with a “fishbowl” networking activityin which students sat in a large circle surrounding a smaller group of four students. Conversations took place within the smaller group, while the large group observed, and all students had a chance to participate in the smaller group discussion. Thus, students were able to introduce themselves to all participants while engaging in focused conversations that allowed them time to describe their research interests and to brainstorm possibilities for collaboration in the area of Big Data. We followed this activity with a networking activity during which students were randomly paired to talk further about their projects and interests. Finally, students had the opportunity to network with other students of their choice, in order to follow up on particularly interesting projects or collaborations they had learned about during the course of the evening. The success of this event far exceeded our expectations—out of the five fellowship-winning teams for this year, at least three are composed of members who met for the first time that evening!

Following the Big Data Mixer, potential collaborators were invited to submit short, two-page “pre-proposals,”which were due in mid-January, in order to get feedback on their fellowship project ideas in advance of the longer proposal for the final fellowship competition. As an incentive, the team members for the top six pre-proposals were offered $500 in travel funds. In total, 12 pre-proposals were submitted. All teams who submitted pre-proposals subsequently received comments and suggestions for improvement from the judging committee. While pre-proposals provided teams with a valuable opportunity to receive feedback on their work, and most teams whose projects eventually were awarded fellowships did submit pre-proposals, the final application process was open to all applicants, whether they had or had not submitted pre-proposals.

Thirteen teams (including six that we had not seen at the pre-proposal stage) submitted fellowship applications in February 2014, including teams of students working together from Politics and Systems Engineering; Psychology and Computer Science; and Nursing and Biomedical Engineering. The projects chosen for funding during the 2014-15 academic year are as follows:

  • Gut Predictions: Building a Predictive Model of the Gut Biome: Matthew Biggs, Biomedical Engineering, and Steven Steinway, School of Medicine
  • Data-Driven, Complex Systems Model of Public Opinion: Osama Eshera and Gerard Learmonth, Systems and Information Engineering, and Chelsea Goforth and Nicholas Winter, Politics
  • Big Data Meets Mind-Body Complementary Therapy: Data-Driven Musculoskeletal Models that Guide the Use of Yoga for Treatment of Chronic Diseases: Tamara G. Fischer-White, Nursing, and Kelley Virgilio, Biomedical Engineering
  • Uncovering Mechanisms of Social Movements within the Arab Spring: A Data-Mining Approach, Robert Kubinec, Politics, and Congyu Wu, Systems Engineering
  • Maintained Individual Data Distributed Likelihood Estimation (MIDDLE): Joshua N. Pritikin, Psychology, and Yang Wang, Systems and Information Engineering

These fellowships, and the process leading up to the fellowship proposals, make a crucial contribution to the goals of our office and to the quality of graduate student research at the University of Virginia. Students who otherwise never would have known about each other’s research have begun collaborations that will enhance the individual work of each student and, most importantly, will bring to contemporary problems – problems which, of course, do not fit neatly into disciplinary categories – perspectives and solutions that would not be possible without such work. This has the potential to improve research across the University, as these students bring their experiences to bear on their work with other graduate students, faculty, and undergraduates, and it allows our brightest young researchers to apply their skills and knowledge to the most pressing contemporary questions, both intellectual and practical, that face human communities near and far.